In the 2016 Presidential Election, Donald Trump won the Rust Belt, where middle class jobs in manufacturing had vanished. That “restoring the middle class” has made it to the top of the policy agenda in Washington and elsewhere is thus understandable. Now the crucial question is how it can be achieved.
Since taking office, President Joe Biden has pledged to deliver for the middle class by promoting labour unions and lifting the federal minimum wage to $15. He has also pushed for an Infrastructure Investment and Jobs Act, which is yet to pass the House, and would funnel billions into building bridges, highways and other critical infrastructure, adding perhaps thousands of jobs in construction.
But while a raising the minimum wage could make existing jobs less bad, it won’t restore the middle class. Nor would much needed infrastructure spending, which is essentially a one-off.
Old middle class jobs aren’t coming back. They have been permanently lost to the process of technological change. Even in China, over 12 million manufacturing jobs have been shredded only in the past few years.
The middle class can be restored if, and only if, new middle class jobs are created. And that needs more innovation. As research by David Autor, Anna Salomons, and Bryan Seegmiller shows, the majority of today’s jobs didn’t even exist in 1940. Like the jobs of solar panel installers and software engineers, they had to be invented.
The economic case for governments to invest in innovation is straightforward: the socially optimal level of R&D investment—which produces the highest rate of economic growth—is much lower than actual spending.
But more spending alone won’t be enough. To create jobs, new technological knowledge must be absorbed and translated into new prototypes, products, and industries. And this is best done by start-ups.
While big business excels at incremental improvements of existing technologies, and improving efficiency through automation, this if anything shreds jobs. Small and young companies, in contrast, have a comparative advantage in radical innovation, as Ufuk Akcigit and William Kerr document, and are thus more likely to create the industries of the future.
One reason is organizational. Independent inventors and new companies are not tamed by rigid hierarchies and embedded group think. When Lingfei Wu and coauthors analyzed more than 65 million papers, patents and software projects, they found that solo-inventors and smaller teams overwhelmingly stand for the most foundational discoveries.
Another reason is strategic. When new technologies threaten revenues from old ones, incumbents are naturally less likely to pursue them. Kodak, which developed the first digital camera in 1975, but then dropped the product, is a classic example.
But much more worrying, incumbents also try to shield themselves from competition. In a recent study, Thomas Philippon and Germán Gutiérrez found that “regulations have a negative impact on small firms, especially in industries with high lobbying expenditures.”
Thanks to the Web, open-source software, and cloud computing, the cost of starting a new company has fallen dramatically in recent years—from around $5 million a decade ago to less than $50,000 today by some estimates. Yet despite this, new firm formation in the US tech sector has actually declined since the 2000s.
Thus, competition policy has an important role to play beyond increasing the bargaining power of workers by making them less dependent on a few employers. It could also facilitate the creation of entirely new types of work. The 1956 consent decree, for example, which forced Bell Labs to licence its patents, led to a lasting increase in innovation.
This is not to suggest that restoring competition is a panacea for all ills of the middle class. As I have argued in my book The Technology Trap, right now, many start-ups are inventing technologies that replace rather than create jobs. The Covid-19 pandemic has led to an upsurge in start-ups focusing on restaurant automation, to take just one example.
Yet the direction of technological change isn’t destiny. Currently, low taxes on capital and high taxes on labor favors automation at the expense of job-creating technologies. As Daron Acemoglu, Andrea Manera and Pascual Restrepo document, eliminating the capital bias in the tax code could boost employment as well as the share of national income going to labor.
What’s more, much more can be done to increase the pool of potential innovators and entrepreneurs. We know, for example, that children who are exposed to innovation early in life—often through their parents—are more much likely to become inventors themselves. Thus, leveraging the educational system to expose more children to innovation in the formative school years seems like low-hanging fruit.
Not only would there be fewer lost Einsteins. It would mean more people inventing the jobs of the future.